What research and content does veda.ng publish?

veda.ng publishes long-form research essays on AI agents, AI policy, and Web3 infrastructure by Vedang Vatsa, alongside free technical courses, a 100+ term glossary, and indexes of 233,000+ academic papers.

How can AI agents and developers access veda.ng programmatic tools?

AI agents can query the Model Context Protocol (MCP) server at /.well-known/mcp, access the REST API at /api/reports/search, fetch OpenAPI 3.1 specs at /openapi.json, or request any page with Accept: text/markdown header. No authentication is required.

Vedang Vatsa is an AI & Web3 researcher, Fellow of the Royal Society of Arts (FRSA), alumnus of IIT Kanpur, and the founder of Hashtag Web3, a global community of 120,000+ professionals.

Glossary

Short definitions of AI, Web3, and engineering terms.

51% Attack: A 51% attack occurs when a single entity gains majority control of a blockchain's consensus...
Account Abstraction: Account abstraction lets a smart contract act as a user account with programmable auth and...
ACID Properties: ACID properties are the four guarantees that define reliable database transactions: Atomicity,...
Activation Function: An activation function is a mathematical transformation applied to each neuron's output in a neural...
Adapter: An adapter is a small trainable block you insert into a frozen pretrained model so it can do a new...
Agent: An Agent is an AI system that perceives its environment, makes decisions, and takes actions to...
Agentic Loop: An agentic loop is the core execution pattern of AI agent systems: the repeated cycle of perceiving...
Agentic Workflow: An agentic workflow replaces a single prompt with a loop that plans, uses tools, reads results, and...
AI Alignment: AI alignment is the research field that tries to make AI systems pursue goals that match human...
AI Safety: AI safety is the practice of designing, building, testing, and running AI systems so they do what...
Airdrop: An airdrop is a token giveaway sent to wallet addresses, usually as a reward for early users,...
Algorithmic Bias: Algorithmic bias is when an automated decision process produces systematically unfair outcomes for...
Alignment: Alignment is the problem of making an AI system pursue goals that match human values. People...
API (Application Programming Interface): An API is an interface that lets one program talk to another without opening its files or database....
APY vs APR: APR, Annual Percentage Rate, and APY, Annual Percentage Yield, both describe yearly interest. APR...
Artificial General Intelligence (AGI): Artificial General Intelligence is a system that matches or exceeds human-level performance across...
Artificial Superintelligence (ASI): Artificial Superintelligence, ASI, is a hypothetical machine intelligence that would beat humans...
Atomic Swap: An atomic swap is a trustless peer-to-peer exchange of coins across different chains using hash...
Attention Head: An attention head is one of multiple parallel attention mechanisms within a transformer layer, each...
Attention Mechanism: The attention mechanism is the core innovation inside transformer models that allows them to weigh...
Attention Score: An attention score quantifies how much one position in a sequence should attend to another position...
Autoencoder: An autoencoder is a neural network trained to reconstruct its input through a compressed...
Automated Market Maker (AMM): An automated market maker is a DEX design that prices trades with a formula and a liquidity pool,...
Autonomous Agents: An autonomous agent is an AI system that pursues a multi-step goal without waiting for a click or a...
Backpropagation: Backpropagation is the algorithm that makes deep learning possible by efficiently computing how...
Basis Trade: Basis trade profits from the price difference between spot and futures markets, exploiting the...
Batch Normalization: Batch normalization is a technique for stabilizing and accelerating neural network training by...
Beam Search: Beam search is a decoding algorithm for language models that maintains multiple candidate sequences...
BERT: BERT (Bidirectional Encoder Representations from Transformers) is a transformer-based language...
Blockchain: A blockchain is a distributed ledger that a network of nodes maintains together. Each block stores...
Borrow Rate: Borrow rate is the interest rate paid to borrow assets from a DeFi lending protocol, dynamically...
Bridge: A bridge is a protocol for moving assets between blockchains. Chains are isolated. Bitcoin cannot...
Byte-Pair Encoding: Byte-pair encoding builds a subword vocabulary by merging the most frequent adjacent pairs again...
Caching Strategy: Caching strategies determine how data flows between cache and underlying storage, optimizing for...
Causal Masking: Causal masking blocks each position from attending to later tokens so generation can only use left...
CDN (Content Delivery Network): A Content Delivery Network is a group of servers spread across many locations around the world that...
Chain-of-Thought Prompting: Chain-of-thought prompting is a technique where you instruct a language model to reason step by...
Chunking: Chunking splits documents into smaller segments for embedding and retrieval in RAG systems. The...
CI/CD: CI/CD is Continuous Integration and Continuous Deployment: test every commit, then ship every build...
Circuit Breaker: A circuit breaker is a software design pattern that prevents cascading failures in distributed...
Circulating Supply: Circulating supply is the number of tokens actually available to trade, excluding locked...
CLIP: CLIP is an OpenAI model that maps images and text into one embedding space using contrastive...
Cluster Analysis: Cluster analysis groups items so members of a group are more similar to each other than to members...
Cold Storage: Cold storage means keeping private keys on a device with no internet. Remote hacks then have...
Cold Wallet: A cold wallet is a cryptocurrency storage method where private keys are kept offline, completely...
Collateralization Ratio: Collateralization ratio is the value of posted collateral divided by borrowed assets, as a...
Composability: Composability in DeFi is the ability of smart contracts to call each other in one atomic...
Computer Vision: Computer vision is the AI field that interprets images and video, turning pixels into structured...
Concentrated Liquidity: Concentrated liquidity lets providers choose a price range for their capital instead of spreading...
Consensus Mechanism: A consensus mechanism is a protocol that lets nodes agree on blockchain state without a central...
Consistency Hashing: Consistent hashing maps keys onto a ring of servers so adding or removing a node moves only a slice...
Constitutional AI: Constitutional AI is a training approach where an AI system evaluates its own outputs against a set...
Containerization (Docker): Containerization packs an app with its runtime so it starts the same on any machine that has the...
Context Window: A context window is the maximum amount of text a language model can process in a single...
Continuous Integration / Continuous Deployment (CI/CD): Continuous Integration / Continuous Deployment (CI/CD) automates the pipeline from writing code to...
Contrastive Learning: Contrastive learning pulls similar items together and pushes dissimilar items apart in embedding...
Convolutional Neural Network: A convolutional neural network is a neural architecture designed for processing grid-structured...
CORS: Cross-Origin Resource Sharing (CORS) is a browser security mechanism that controls which web pages...
Cross-Attention: Cross-attention is an attention mechanism where one sequence attends to a different sequence,...
Cross-Chain Bridge: A cross-chain bridge moves tokens, data, or messages between blockchains that cannot talk to each...
Cross-Entropy Loss: Cross-entropy loss is the standard objective function for training classification and language...
Cryptocurrency Wallet: A cryptocurrency wallet stores the private and public keys used to send, receive, and manage assets...
Cryptographic Hash Function: A cryptographic hash function takes an input of any size and produces a fixed-length string of...
Data Availability: Data availability (DA) guarantees that transaction data is accessible to anyone who needs to verify...
Data Availability Sampling (DAS): Data Availability Sampling (DAS) is a protocol that lets nodes verify a block's data is fully...
Database Index: A database index is a data structure that accelerates query performance by maintaining sorted...
Database Sharding: Database sharding horizontally partitions data across multiple database instances based on a shard...
Decentralized Autonomous Organization (DAO): A DAO is an organization run by smart contracts and token holders instead of a CEO and a board....
Decentralized Exchange (DEX): A decentralized exchange, DEX, lets users trade through smart contracts without a company holding...
Decentralized Finance (DeFi): Decentralized finance is financial services on a blockchain without a bank, broker, or exchange in...
Decentralized Identity (DID): Decentralized Identity (DID) is an architectural change in how identity, credentials, and...
Decentralized Physical Infrastructure Networks (DePIN): Decentralized Physical Infrastructure Networks (DePIN) combine real-world hardware assets, like...
Decentralized Sequencer: A decentralized sequencer orders transactions in a distributed ledger without relying on a single...
Deep Learning: Deep learning is a subfield of machine learning that uses neural networks with many layers to learn...
Delegated Proof of Stake (DPoS): Delegated Proof of Stake, DPoS, is consensus where token holders elect a small set of delegates or...
Delta Neutral: Delta neutral means a book with no net bet on price direction, so you try to earn other yields....
DePIN: Decentralized Physical Infrastructure Networks are blockchain projects that pay people in tokens to...
Diffusion Model: A diffusion model makes data by learning to undo noise. Training adds Gaussian noise to real...
Digital Twin: A digital twin is a virtual replica of a physical object, process, or system that updates in real...
Direct Preference Optimization: Direct Preference Optimization fine-tunes a language model on preference pairs without training a...
Directed Acyclic Graph (DAG): A Directed Acyclic Graph, DAG, is a set of nodes and directed edges with no cycles. Follow edges...
DNS (Domain Name System): The Domain Name System maps a hostname such as google.com to an IP address such as 142.250.80.46....
Docker: Docker packages an application and its dependencies into a container image you can run the same way...
Document Retrieval: Document retrieval finds relevant documents for a query. It is the first stage of search engines...
Double Spend: Double spending is using the same digital token twice before the network treats the first spend as...
Double-Spending: Double-spending is spending the same digital coin more than once. Cash changes hands. Files copy....
Dropout: Dropout is a regularization technique that randomly deactivates a fraction of neurons during each...
Dust Attack: A dust attack is a privacy break that sends tiny amounts of crypto, dust, to many addresses, then...
Edge AI: Edge AI runs AI models on devices at the network edge, phones, industrial sensors, cameras, drones,...
Edge Computing: Edge computing processes data near the device or user instead of sending every byte to a distant...
Embeddings: Embeddings are vector representations of text, images, or other data in high-dimensional space....
Emission Rate: Emission rate measures the speed at which new tokens enter circulation, typically expressed as...
Ethereum Virtual Machine (EVM): The Ethereum Virtual Machine, EVM, is the shared computer that runs Ethereum smart contracts....
Expert Systems: An expert system is a program that copies a human specialist's decision process in one domain. It...
Explainable AI (XAI): Explainable AI, XAI, is a set of techniques and design rules that make machine-learning decisions...
Feature Extraction: Feature extraction turns raw data into a smaller set of numbers that algorithms can use for...
Federated Learning: Federated learning trains a model on many devices without collecting the raw data in one place. The...
Feedforward Network: A feedforward network in a transformer is a two-layer net applied independently to each position...
Few-Shot Prompting: Few-shot prompting is a technique where you provide a small number of input-output examples before...
Fine-Tuning: Fine-tuning is the process of taking a pre-trained model and continuing its training on a...
Flash Loan: A flash loan is an uncollateralized DeFi loan that must be borrowed and repaid inside one...
Foundation Model: A foundation model is a large AI model trained on broad data at massive scale that serves as the...
Funding Rate: Funding rate is a periodic payment between long and short perpetual holders that pulls the...
Gas Fees: Gas fees are transaction fees paid to validators for running operations on a blockchain. The name...
Generalization: Generalization is whether a model does well on new data from the same distribution as the training...
Generative Adversarial Network: A generative adversarial network trains two neural networks in competition: a generator that...
Generative AI: Generative AI is the class of machine-learning models that create new text, images, audio, video,...
Generative Model: A generative model learns the probability distribution of training data, enabling it to create new...
Genesis Block: The genesis block is the first block of a blockchain. Every later block hashes back to it. It has...
Git: Git is a distributed version control system that tracks changes to files over time, letting...
Governance Token: A governance token is a digital asset that lets you vote in a DAO. Holders vote on protocol...
Gradient Descent: Gradient descent trains a neural network by moving the weights a little, again and again, in the...
GraphQL: GraphQL is a query language for APIs where the client names the fields and the server returns that...
Grounding: Grounding in AI refers to connecting model outputs back to verifiable external reality, so claims...
gRPC: gRPC is a high-performance, open-source remote procedure call (RPC) framework that uses Protocol...
Hallucination: Hallucination is when an AI system generates plausible-sounding but completely false information....
Hard Fork: A hard fork is a blockchain protocol upgrade that is not backwards-compatible. Nodes running the...
Hot Wallet: A hot wallet keeps keys on an internet-connected device so you can transact immediately. That...
Hyperparameter: A hyperparameter is a setting you choose before training. It controls learning. It is not a weight...
Hyperparameter Tuning: Hyperparameter tuning is choosing the settings that control how a machine-learning model learns,...
Idempotency: Idempotency is the property where performing an operation multiple times produces the same result...
Image Segmentation: Image segmentation assigns a class label to every pixel, giving dense, pixel-level understanding...
Impermanent Loss: Impermanent loss is the gap liquidity providers take when the price ratio of paired assets moves...
Inference: Inference is the process of running a trained AI model to generate predictions or outputs. It is...
Instruction Tuning: Instruction tuning fine-tunes language models on many instruction-response pairs so the model...
Intent: An intent is a signed message expressing what a user wants to achieve without specifying exactly...
Intent-Centric Architecture: Intent-Centric Architecture is a design approach where the user's desired outcome, called the...
Internet of Things (IoT): The Internet of Things (IoT) refers to a network of physical objects, sensors, appliances,...
IPFS (InterPlanetary File System): IPFS is a peer-to-peer protocol for storing and sharing files. HTTP is location-addressed. You ask...
Knowledge Distillation: Knowledge distillation is a model compression technique where a smaller student model learns to...
Knowledge Graph: A knowledge graph stores facts as entities connected by typed relationships. People, places,...
Kubernetes: Kubernetes is a platform that deploys, scales, and heals containerized applications across a...
KV Cache: KV cache stores Key and Value vectors from earlier tokens during generation so they are not...
Large Language Model (LLM): A Large Language Model is a neural network trained on massive text datasets to predict and generate...
Latency: Latency measures the time delay between initiating a request and receiving a response, a critical...
Layer 1: Layer 1 is the base blockchain. It provides security, consensus, and settlement finality. Bitcoin...
Layer 2: Layer 2 means secondary chains built on a Layer 1 such as Ethereum to raise throughput. Ethereum...
Layer Normalization: Layer normalization stabilizes training by normalizing each sample across features so mean and...
Leverage: Leverage raises exposure above your cash. Gains and losses scale by the factor. At 10x, $100...
Limit Order: A limit order names the exact price you will trade at. You get price control and you may not fill....
Liquid Restaking: Liquid Restaking takes the capital efficiency of liquid staking to its logical and arguably risky...
Liquid Staking: Liquid staking lets you stake tokens and still use a liquid stand-in. Normal staking locks the...
Liquidation: Liquidation in DeFi is forced closure of an undercollateralized loan. The protocol sells collateral...
Liquidity Pool: A liquidity pool is a smart contract that holds reserves of two or more tokens so people can trade...
Liquidity Pool Volume: Liquidity pool volume is the dollar value of swaps through a pool over a period, usually 24 hours....
Liquidity Provider: A liquidity provider, LP, deposits assets into pools or order books so others can trade. LPs earn...
Load Balancer: A load balancer is a system that distributes incoming network traffic across multiple servers to...
Logit: Logits are the raw, unnormalized scores from a neural network's final layer, before softmax turns...
LoRA: LoRA, Low-Rank Adaptation, adds small trainable low-rank matrices to frozen weights for cheap task...
Low-Rank Adaptation (LoRA): Low-Rank Adaptation fine-tunes large nets cheaply by training two small matrices instead of every...
LSTM: Long Short-Term Memory is a recurrent network that can keep information across a long sequence. It...
Machine Translation: Machine Translation, MT, automatically translates text or speech between languages. It moved from...
Margin: Margin is collateral deposited to open and maintain a leveraged trading position, serving as a...
Market Cap: Market capitalization represents the total value of a cryptocurrency, calculated by multiplying the...
Market Order: A market order fills immediately at the best available price. Speed comes first. Price is whatever...
Maximal Extractable Value (MEV): Maximal Extractable Value is profit a block producer can take by choosing the order of...
Merkle Tree: A Merkle tree is a hash tree that proves one record belongs to a large set without downloading the...
Message Queue: A message queue is an asynchronous communication pattern where services send messages to a queue...
MEV Burn: MEV burn is a proposed design that would destroy Maximal Extractable Value instead of paying it to...
Microservices: Microservices is an architecture that splits an app into independent services, each owning one...
Mining Pool: A mining pool is a group of miners who combine hash power, share work, and split block rewards....
Mixture of Experts (MoE): Mixture of Experts is a neural network architecture where only a fraction of the model's parameters...
Model Context Protocol (MCP): Model Context Protocol is an open standard developed by Anthropic that defines how AI models...
Model Distillation: Model distillation is a technique for creating a smaller, faster model that approximates the...
Model Quantization: Model quantization stores weights in lower precision, such as 8-bit integers instead of 32-bit...
Modular Blockchain: A modular blockchain abandons the monolithic architecture where a single network attempts to handle...
Monorepo: A monorepo keeps many projects, packages, or services in one version-control repository instead of...
Multi-Agent Systems: Multi-Agent Systems, MAS, are groups of autonomous software entities that share an environment and...
Multimodal AI: Multimodal AI is a category of models that process multiple data types, text, images, audio, video,...
Named Entity Recognition: Named Entity Recognition finds and classifies people, organizations, locations, dates, money, and...
Neural Network: A neural network is a computational system loosely inspired by biological neurons, consisting of...
Non-Fungible Token (NFT): An NFT is a unique cryptographic token that records ownership of a specific digital or physical...
OAuth: OAuth is an authorization framework that allows third-party applications to access user resources...
Object Detection: Object detection is a computer vision task that both labels objects in an image and locates them...
Optical Character Recognition: OCR converts images of text, scans, signs, and handwriting into machine-readable characters. Older...
Optimistic Rollup: Optimistic rollups treat transactions as valid by default and post state to Layer 1 without an...
Oracle: An oracle is a service that brings off-chain data to smart contracts. Blockchains are closed...
Order Book: An order book is a live list of buy and sell orders for an asset, with size at each price. Bids are...
Overfitting: Overfitting is when a model memorizes the training set instead of learning patterns that hold on...
Parallelized EVM: A parallelized EVM runs Ethereum contract execution across CPU cores instead of one instruction at...
Pattern Recognition: Pattern recognition finds regularities in data, from faces in a crowd to models sorting spam from...
Payment Channel: A payment channel is a Layer 2 method that lets two parties make unlimited instant payments by...
Peer-to-Peer (P2P): Peer-to-Peer (P2P) is a network architecture where each node acts as both client and server,...
Perpetual Futures: Perpetual futures are derivatives that track an asset's price with no expiry. BitMEX popularized...
Perplexity: Perplexity is a metric measuring how well a language model predicts a test dataset, calculated as...
Perplexity Trap: The perplexity trap is the dangerous assumption that lower perplexity on benchmark datasets...
Positional Encoding: Positional encoding is a technique for injecting sequence position information into transformer...
Predictive Analytics: Predictive analytics turns past and present data into forecasts using statistics, data mining, and...
Prefix Tuning: Prefix tuning adapts a frozen language model by learning continuous vectors prepended to the input....
Prompt Engineering: Prompt engineering is the craft of designing inputs to AI systems to produce desired outputs. It...
Prompt Injection: Prompt injection is a security attack where malicious instructions are embedded in content that an...
Prompt Tuning: Prompt tuning learns continuous soft prompt embeddings prepended to inputs while all model weights...
Proof of History (PoH): Proof of History, PoH, is a cryptographic clock that orders events without nodes arguing over...
Proof of Stake: Proof of Stake is consensus where validators propose and attest to blocks based on how much...
Proof of Work: Proof of Work is the original blockchain consensus method, used by Bitcoin. Participants spend...
Protocol Buffer: Protocol Buffers (protobuf) is a language-neutral, platform-neutral data serialization format...
Protocol Revenue: Protocol revenue is the income a DeFi protocol generates from fees and services, representing...
Public-Key Cryptography: Public-key cryptography uses two linked keys: a public key anyone can see and a private key only...
Quantization: Quantization is a model compression technique that reduces the precision of a neural network's...
Quantum Computing: Quantum computing uses qubits and entanglement to run algorithms that do not map cleanly onto...
Question Answering: Question Answering systems answer natural language questions by extracting a span or generating a...
Random Forest: Random Forest is an ensemble that trains many decision trees and combines their predictions. Each...
Rate Limiting: Rate limiting is controlling the number of requests a client can make to a service in a time...
Re-ranking: Re-ranking is a two-stage retrieval pattern: a fast retriever proposes candidates, then a slower,...
Real World Assets (RWA): Real World Assets, RWAs, are tokens that represent off-chain assets such as real estate, treasury...
Reasoning Model: A reasoning model spends time thinking before it answers. It uses an internal chain of thought to...
Recurrent Neural Network: A recurrent neural network processes sequential data by maintaining a hidden state that carries...
Reinforcement Learning: Reinforcement Learning trains an agent through actions and numeric rewards rather than a labeled...
Reinforcement Learning from Human Feedback (RLHF): Reinforcement Learning from Human Feedback is a training technique where humans rank model outputs...
Residual Connection: A residual connection, also called a skip connection, adds a layer's input directly to its output...
REST API: A REST API is an HTTP interface where each resource has a URL and clients act on it with standard...
Restaking: Restaking lets ETH stakers reuse staked ETH as security for protocols beyond Ethereum. Normally,...
Retrieval-Augmented Generation (RAG): Retrieval-Augmented Generation is a technique where an LLM queries external knowledge bases before...
Reward Model: A reward model is a neural network trained to predict human preferences, serving as the learned...
Rollup: A rollup is a Layer 2 that executes transactions off-chain for speed and cost, then posts data or...
Rug Pull: A rug pull is a crypto scam where the team takes the money and leaves holders with a dead token....
SAM: SAM is a Meta AI segmentation model that can mask objects from a point, box, or text prompt without...
Scaffold: A scaffold in AI prompting is a structured template that guides models through complex reasoning by...
Scaling Law: Scaling laws are empirical relationships that predict how neural network performance improves as...
Seed Phrase: A seed phrase, also called a recovery phrase or mnemonic phrase, is a sequence of 12 or 24 randomly...
Self-Attention: Self-attention is an attention mechanism where a sequence attends to itself, allowing each position...
Self-Sovereign Identity (SSI): Self-sovereign identity is a model where you hold your own credentials instead of renting an...
Semantic Similarity: Semantic similarity scores how close two texts are in meaning, not just shared words, for duplicate...
Semi-Supervised Learning: Semi-supervised learning builds models from both labeled and unlabeled data. Fully supervised...
Sentiment Analysis: Sentiment analysis scores the opinion or emotional tone in text, usually as positive, negative, or...
Sequencer: A sequencer orders and batches rollup transactions before posting them to Layer 1. It decides...
Serverless: Serverless is a cloud model where the provider runs the machines. You write functions. They scale...
Serverless Architecture: Serverless architecture means you deploy functions and the cloud runs them on events, billed by...
Sharding: Sharding is a horizontal scaling technique that partitions data across multiple independent...
Slashing: Slashing is an automatic penalty in Proof of Stake. If a validator breaks the protocol, the chain...
Slippage: Slippage is the gap between the price you saw when you started a trade and the price you got. It is...
Smart Contract: A smart contract is self-executing code deployed on a blockchain. It enforces an agreement when the...
Social Recovery: Social recovery restores a wallet through trusted contacts, guardians, instead of a seed phrase....
Soft Fork: A soft fork tightens rules in a way old software still accepts. New blocks remain valid to old...
Softmax: Softmax is a mathematical function that transforms a vector of arbitrary real numbers into a...
Soulbound Token (SBT): Soulbound Tokens (SBTs) are non-transferable, identity-centric digital tokens that uniquely...
Sparse Expert: A sparse expert is one specialist network inside a Mixture of Experts model. Only a few experts run...
Spatial Computing: Spatial computing maps a room and places digital objects in that geometry so software can sit on...
Speculative Decoding: Speculative decoding lets a small draft model propose tokens that a larger target model verifies in...
Speech Recognition: Speech recognition, or ASR, converts spoken language into text for voice interfaces, transcripts,...
SSL/TLS: SSL (Secure Sockets Layer) and TLS (Transport Layer Security) are cryptographic protocols that...
Stablecoin: A stablecoin is a cryptocurrency pegged to a stable asset such as the US dollar. Bitcoin moved from...
State Channel: A state channel lets parties update state off-chain with on-chain security, settling on-chain only...
Supervised Learning: Supervised learning trains a model on labeled examples where every input has a known correct...
Supply Rate: Supply rate is the interest lenders earn for depositing into a DeFi lending protocol. It comes from...
Support Vector Machine (SVM): Support Vector Machine finds the widest-margin hyperplane that separates classes, for...
Sybil Attack: A Sybil attack is when one adversary creates thousands or millions of fake identities to grab power...
Synthetic Data: Synthetic data is artificially generated data created to train, test, or evaluate AI systems, as...
TCP/IP: TCP/IP is the protocol pair that addresses hosts, packs data into packets, routes them, and...
Temperature: Temperature is a parameter that controls how random language model output is by scaling the logits...
Text-to-Speech: Text-to-Speech, TTS, synthesizes spoken audio from written text for voice interfaces, audiobooks,...
Throughput: Throughput measures the rate at which a system processes work over time, typically expressed as...
Time-Weighted Average Price: Time-Weighted Average Price, TWAP, is the average price of an asset over a window, weighted by how...
Token: A token is the smallest unit of text that an LLM processes. It's not a word. A word is often...
Token Burn: Token burning permanently removes tokens by sending them to an address with no private key, a burn...
Tokenization: Tokenization in Web3 means representing ownership of a real-world or digital asset as a token on a...
Tokenizer: A tokenizer converts raw text into a sequence of discrete tokens a language model can process,...
Tokenomics: Tokenomics is the economic design of a cryptocurrency or token system, the rules governing how...
Top-k Sampling: Top-k sampling limits the next token to the k most probable items so rare tokens are less likely to...
Top-p Sampling: Top-p sampling, or nucleus sampling, keeps the smallest token set whose probabilities add up to at...
Total Value Locked: Total Value Locked, TVL, is the dollar value of assets deposited in a DeFi protocol. People use it...
Transfer Learning: Transfer learning is a machine learning approach where knowledge gained from training on one task...
Transformer: The Transformer is a neural network architecture introduced in the 2017 paper "Attention Is All You...
Transformer Architecture: The Transformer, introduced in 2017, processes sequences with attention instead of recurrent...
Utilization Rate: Utilization rate is the share of deposited assets currently borrowed. If $80 million is borrowed...
Validator: A validator is a node that proposes and checks new blocks on a Proof of Stake chain. In Proof of...
Value at Risk: Value at Risk, VaR, is the maximum expected loss on a portfolio over a time horizon at a stated...
Variational Autoencoder: A variational autoencoder is a generative model that learns a probabilistic latent space where...
Vector Database: A vector database is a specialized database optimized for storing and searching high-dimensional...
Vesting Schedule: A vesting schedule releases tokens to recipients gradually over time rather than all at once,...
Vision Transformer: Vision Transformer, ViT, splits an image into patches, treats each patch as a token, and runs a...
Vision Transformers (ViT): Vision Transformers, ViT, adapt the transformer, first built for text, to images. Instead of...
Vocabulary: In a language model, the vocabulary is the full set of tokens the model knows. Each token has an...
Wallet: A wallet is software that stores private keys and lets you use a blockchain. The private key is the...
Web2: Web2 is the read-write web that took over in the early 2000s: sites as platforms and users as...
WebAssembly (Wasm): WebAssembly is a binary instruction format for a stack machine that runs at near-native speed in...
Webhook: A webhook is an HTTP callback: the source POSTs to your URL when an event happens, instead of you...
WebSocket: WebSocket is a communication protocol that provides full-duplex, persistent connections between...
Word Embeddings: Word embeddings are numeric vectors for words that place meaning and relationships in a continuous...
Wrapped Token: A wrapped token is a cryptocurrency that stands in for another asset on a different chain, pegged...
Yield: Yield in DeFi is the return from putting capital into protocols, usually quoted as APY, the...
Yield Farming: Yield farming is deploying crypto across DeFi protocols to stack trading fees, interest, and token...
Zero-Knowledge EVM (zkEVM): A zkEVM is a zero-knowledge rollup that can prove standard Ethereum contract execution. A ZK-Rollup...
Zero-Knowledge Proof: A zero-knowledge proof is a cryptographic method to prove knowledge of information without...
Zero-Knowledge Proofs (ZKPs): Zero-knowledge proofs (ZKPs) are cryptographic protocols that let one party (the prover) convince...
Zero-Shot Learning: Zero-shot learning is when an AI system performs tasks it was never explicitly trained on, using...
ZK Rollup: ZK rollups post cryptographic validity proofs to Layer 1 that the batch is correct, so they skip...

Showing 287 of 287 terms

Glossary

Short definitions of AI, Web3, and engineering terms.

51% Attack: A 51% attack occurs when a single entity gains majority control of a blockchain's consensus...
Account Abstraction: Account abstraction lets a smart contract act as a user account with programmable auth and...
ACID Properties: ACID properties are the four guarantees that define reliable database transactions: Atomicity,...
Activation Function: An activation function is a mathematical transformation applied to each neuron's output in a neural...
Adapter: An adapter is a small trainable block you insert into a frozen pretrained model so it can do a new...
Agent: An Agent is an AI system that perceives its environment, makes decisions, and takes actions to...
Agentic Loop: An agentic loop is the core execution pattern of AI agent systems: the repeated cycle of perceiving...
Agentic Workflow: An agentic workflow replaces a single prompt with a loop that plans, uses tools, reads results, and...
AI Alignment: AI alignment is the research field that tries to make AI systems pursue goals that match human...
AI Safety: AI safety is the practice of designing, building, testing, and running AI systems so they do what...
Airdrop: An airdrop is a token giveaway sent to wallet addresses, usually as a reward for early users,...
Algorithmic Bias: Algorithmic bias is when an automated decision process produces systematically unfair outcomes for...
Alignment: Alignment is the problem of making an AI system pursue goals that match human values. People...
API (Application Programming Interface): An API is an interface that lets one program talk to another without opening its files or database....
APY vs APR: APR, Annual Percentage Rate, and APY, Annual Percentage Yield, both describe yearly interest. APR...
Artificial General Intelligence (AGI): Artificial General Intelligence is a system that matches or exceeds human-level performance across...
Artificial Superintelligence (ASI): Artificial Superintelligence, ASI, is a hypothetical machine intelligence that would beat humans...
Atomic Swap: An atomic swap is a trustless peer-to-peer exchange of coins across different chains using hash...
Attention Head: An attention head is one of multiple parallel attention mechanisms within a transformer layer, each...
Attention Mechanism: The attention mechanism is the core innovation inside transformer models that allows them to weigh...
Attention Score: An attention score quantifies how much one position in a sequence should attend to another position...
Autoencoder: An autoencoder is a neural network trained to reconstruct its input through a compressed...
Automated Market Maker (AMM): An automated market maker is a DEX design that prices trades with a formula and a liquidity pool,...
Autonomous Agents: An autonomous agent is an AI system that pursues a multi-step goal without waiting for a click or a...
Backpropagation: Backpropagation is the algorithm that makes deep learning possible by efficiently computing how...
Basis Trade: Basis trade profits from the price difference between spot and futures markets, exploiting the...
Batch Normalization: Batch normalization is a technique for stabilizing and accelerating neural network training by...
Beam Search: Beam search is a decoding algorithm for language models that maintains multiple candidate sequences...
BERT: BERT (Bidirectional Encoder Representations from Transformers) is a transformer-based language...
Blockchain: A blockchain is a distributed ledger that a network of nodes maintains together. Each block stores...
Borrow Rate: Borrow rate is the interest rate paid to borrow assets from a DeFi lending protocol, dynamically...
Bridge: A bridge is a protocol for moving assets between blockchains. Chains are isolated. Bitcoin cannot...
Byte-Pair Encoding: Byte-pair encoding builds a subword vocabulary by merging the most frequent adjacent pairs again...
Caching Strategy: Caching strategies determine how data flows between cache and underlying storage, optimizing for...
Causal Masking: Causal masking blocks each position from attending to later tokens so generation can only use left...
CDN (Content Delivery Network): A Content Delivery Network is a group of servers spread across many locations around the world that...
Chain-of-Thought Prompting: Chain-of-thought prompting is a technique where you instruct a language model to reason step by...
Chunking: Chunking splits documents into smaller segments for embedding and retrieval in RAG systems. The...
CI/CD: CI/CD is Continuous Integration and Continuous Deployment: test every commit, then ship every build...
Circuit Breaker: A circuit breaker is a software design pattern that prevents cascading failures in distributed...
Circulating Supply: Circulating supply is the number of tokens actually available to trade, excluding locked...
CLIP: CLIP is an OpenAI model that maps images and text into one embedding space using contrastive...
Cluster Analysis: Cluster analysis groups items so members of a group are more similar to each other than to members...
Cold Storage: Cold storage means keeping private keys on a device with no internet. Remote hacks then have...
Cold Wallet: A cold wallet is a cryptocurrency storage method where private keys are kept offline, completely...
Collateralization Ratio: Collateralization ratio is the value of posted collateral divided by borrowed assets, as a...
Composability: Composability in DeFi is the ability of smart contracts to call each other in one atomic...
Computer Vision: Computer vision is the AI field that interprets images and video, turning pixels into structured...
Concentrated Liquidity: Concentrated liquidity lets providers choose a price range for their capital instead of spreading...
Consensus Mechanism: A consensus mechanism is a protocol that lets nodes agree on blockchain state without a central...
Consistency Hashing: Consistent hashing maps keys onto a ring of servers so adding or removing a node moves only a slice...
Constitutional AI: Constitutional AI is a training approach where an AI system evaluates its own outputs against a set...
Containerization (Docker): Containerization packs an app with its runtime so it starts the same on any machine that has the...
Context Window: A context window is the maximum amount of text a language model can process in a single...
Continuous Integration / Continuous Deployment (CI/CD): Continuous Integration / Continuous Deployment (CI/CD) automates the pipeline from writing code to...
Contrastive Learning: Contrastive learning pulls similar items together and pushes dissimilar items apart in embedding...
Convolutional Neural Network: A convolutional neural network is a neural architecture designed for processing grid-structured...
CORS: Cross-Origin Resource Sharing (CORS) is a browser security mechanism that controls which web pages...
Cross-Attention: Cross-attention is an attention mechanism where one sequence attends to a different sequence,...
Cross-Chain Bridge: A cross-chain bridge moves tokens, data, or messages between blockchains that cannot talk to each...
Cross-Entropy Loss: Cross-entropy loss is the standard objective function for training classification and language...
Cryptocurrency Wallet: A cryptocurrency wallet stores the private and public keys used to send, receive, and manage assets...
Cryptographic Hash Function: A cryptographic hash function takes an input of any size and produces a fixed-length string of...
Data Availability: Data availability (DA) guarantees that transaction data is accessible to anyone who needs to verify...
Data Availability Sampling (DAS): Data Availability Sampling (DAS) is a protocol that lets nodes verify a block's data is fully...
Database Index: A database index is a data structure that accelerates query performance by maintaining sorted...
Database Sharding: Database sharding horizontally partitions data across multiple database instances based on a shard...
Decentralized Autonomous Organization (DAO): A DAO is an organization run by smart contracts and token holders instead of a CEO and a board....
Decentralized Exchange (DEX): A decentralized exchange, DEX, lets users trade through smart contracts without a company holding...
Decentralized Finance (DeFi): Decentralized finance is financial services on a blockchain without a bank, broker, or exchange in...
Decentralized Identity (DID): Decentralized Identity (DID) is an architectural change in how identity, credentials, and...
Decentralized Physical Infrastructure Networks (DePIN): Decentralized Physical Infrastructure Networks (DePIN) combine real-world hardware assets, like...
Decentralized Sequencer: A decentralized sequencer orders transactions in a distributed ledger without relying on a single...
Deep Learning: Deep learning is a subfield of machine learning that uses neural networks with many layers to learn...
Delegated Proof of Stake (DPoS): Delegated Proof of Stake, DPoS, is consensus where token holders elect a small set of delegates or...
Delta Neutral: Delta neutral means a book with no net bet on price direction, so you try to earn other yields....
DePIN: Decentralized Physical Infrastructure Networks are blockchain projects that pay people in tokens to...
Diffusion Model: A diffusion model makes data by learning to undo noise. Training adds Gaussian noise to real...
Digital Twin: A digital twin is a virtual replica of a physical object, process, or system that updates in real...
Direct Preference Optimization: Direct Preference Optimization fine-tunes a language model on preference pairs without training a...
Directed Acyclic Graph (DAG): A Directed Acyclic Graph, DAG, is a set of nodes and directed edges with no cycles. Follow edges...
DNS (Domain Name System): The Domain Name System maps a hostname such as google.com to an IP address such as 142.250.80.46....
Docker: Docker packages an application and its dependencies into a container image you can run the same way...
Document Retrieval: Document retrieval finds relevant documents for a query. It is the first stage of search engines...
Double Spend: Double spending is using the same digital token twice before the network treats the first spend as...
Double-Spending: Double-spending is spending the same digital coin more than once. Cash changes hands. Files copy....
Dropout: Dropout is a regularization technique that randomly deactivates a fraction of neurons during each...
Dust Attack: A dust attack is a privacy break that sends tiny amounts of crypto, dust, to many addresses, then...
Edge AI: Edge AI runs AI models on devices at the network edge, phones, industrial sensors, cameras, drones,...
Edge Computing: Edge computing processes data near the device or user instead of sending every byte to a distant...
Embeddings: Embeddings are vector representations of text, images, or other data in high-dimensional space....
Emission Rate: Emission rate measures the speed at which new tokens enter circulation, typically expressed as...
Ethereum Virtual Machine (EVM): The Ethereum Virtual Machine, EVM, is the shared computer that runs Ethereum smart contracts....
Expert Systems: An expert system is a program that copies a human specialist's decision process in one domain. It...
Explainable AI (XAI): Explainable AI, XAI, is a set of techniques and design rules that make machine-learning decisions...
Feature Extraction: Feature extraction turns raw data into a smaller set of numbers that algorithms can use for...
Federated Learning: Federated learning trains a model on many devices without collecting the raw data in one place. The...
Feedforward Network: A feedforward network in a transformer is a two-layer net applied independently to each position...
Few-Shot Prompting: Few-shot prompting is a technique where you provide a small number of input-output examples before...
Fine-Tuning: Fine-tuning is the process of taking a pre-trained model and continuing its training on a...
Flash Loan: A flash loan is an uncollateralized DeFi loan that must be borrowed and repaid inside one...
Foundation Model: A foundation model is a large AI model trained on broad data at massive scale that serves as the...
Funding Rate: Funding rate is a periodic payment between long and short perpetual holders that pulls the...
Gas Fees: Gas fees are transaction fees paid to validators for running operations on a blockchain. The name...
Generalization: Generalization is whether a model does well on new data from the same distribution as the training...
Generative Adversarial Network: A generative adversarial network trains two neural networks in competition: a generator that...
Generative AI: Generative AI is the class of machine-learning models that create new text, images, audio, video,...
Generative Model: A generative model learns the probability distribution of training data, enabling it to create new...
Genesis Block: The genesis block is the first block of a blockchain. Every later block hashes back to it. It has...
Git: Git is a distributed version control system that tracks changes to files over time, letting...
Governance Token: A governance token is a digital asset that lets you vote in a DAO. Holders vote on protocol...
Gradient Descent: Gradient descent trains a neural network by moving the weights a little, again and again, in the...
GraphQL: GraphQL is a query language for APIs where the client names the fields and the server returns that...
Grounding: Grounding in AI refers to connecting model outputs back to verifiable external reality, so claims...
gRPC: gRPC is a high-performance, open-source remote procedure call (RPC) framework that uses Protocol...
Hallucination: Hallucination is when an AI system generates plausible-sounding but completely false information....
Hard Fork: A hard fork is a blockchain protocol upgrade that is not backwards-compatible. Nodes running the...
Hot Wallet: A hot wallet keeps keys on an internet-connected device so you can transact immediately. That...
Hyperparameter: A hyperparameter is a setting you choose before training. It controls learning. It is not a weight...
Hyperparameter Tuning: Hyperparameter tuning is choosing the settings that control how a machine-learning model learns,...
Idempotency: Idempotency is the property where performing an operation multiple times produces the same result...
Image Segmentation: Image segmentation assigns a class label to every pixel, giving dense, pixel-level understanding...
Impermanent Loss: Impermanent loss is the gap liquidity providers take when the price ratio of paired assets moves...
Inference: Inference is the process of running a trained AI model to generate predictions or outputs. It is...
Instruction Tuning: Instruction tuning fine-tunes language models on many instruction-response pairs so the model...
Intent: An intent is a signed message expressing what a user wants to achieve without specifying exactly...
Intent-Centric Architecture: Intent-Centric Architecture is a design approach where the user's desired outcome, called the...
Internet of Things (IoT): The Internet of Things (IoT) refers to a network of physical objects, sensors, appliances,...
IPFS (InterPlanetary File System): IPFS is a peer-to-peer protocol for storing and sharing files. HTTP is location-addressed. You ask...
Knowledge Distillation: Knowledge distillation is a model compression technique where a smaller student model learns to...
Knowledge Graph: A knowledge graph stores facts as entities connected by typed relationships. People, places,...
Kubernetes: Kubernetes is a platform that deploys, scales, and heals containerized applications across a...
KV Cache: KV cache stores Key and Value vectors from earlier tokens during generation so they are not...
Large Language Model (LLM): A Large Language Model is a neural network trained on massive text datasets to predict and generate...
Latency: Latency measures the time delay between initiating a request and receiving a response, a critical...
Layer 1: Layer 1 is the base blockchain. It provides security, consensus, and settlement finality. Bitcoin...
Layer 2: Layer 2 means secondary chains built on a Layer 1 such as Ethereum to raise throughput. Ethereum...
Layer Normalization: Layer normalization stabilizes training by normalizing each sample across features so mean and...
Leverage: Leverage raises exposure above your cash. Gains and losses scale by the factor. At 10x, $100...
Limit Order: A limit order names the exact price you will trade at. You get price control and you may not fill....
Liquid Restaking: Liquid Restaking takes the capital efficiency of liquid staking to its logical and arguably risky...
Liquid Staking: Liquid staking lets you stake tokens and still use a liquid stand-in. Normal staking locks the...
Liquidation: Liquidation in DeFi is forced closure of an undercollateralized loan. The protocol sells collateral...
Liquidity Pool: A liquidity pool is a smart contract that holds reserves of two or more tokens so people can trade...
Liquidity Pool Volume: Liquidity pool volume is the dollar value of swaps through a pool over a period, usually 24 hours....
Liquidity Provider: A liquidity provider, LP, deposits assets into pools or order books so others can trade. LPs earn...
Load Balancer: A load balancer is a system that distributes incoming network traffic across multiple servers to...
Logit: Logits are the raw, unnormalized scores from a neural network's final layer, before softmax turns...
LoRA: LoRA, Low-Rank Adaptation, adds small trainable low-rank matrices to frozen weights for cheap task...
Low-Rank Adaptation (LoRA): Low-Rank Adaptation fine-tunes large nets cheaply by training two small matrices instead of every...
LSTM: Long Short-Term Memory is a recurrent network that can keep information across a long sequence. It...
Machine Translation: Machine Translation, MT, automatically translates text or speech between languages. It moved from...
Margin: Margin is collateral deposited to open and maintain a leveraged trading position, serving as a...
Market Cap: Market capitalization represents the total value of a cryptocurrency, calculated by multiplying the...
Market Order: A market order fills immediately at the best available price. Speed comes first. Price is whatever...
Maximal Extractable Value (MEV): Maximal Extractable Value is profit a block producer can take by choosing the order of...
Merkle Tree: A Merkle tree is a hash tree that proves one record belongs to a large set without downloading the...
Message Queue: A message queue is an asynchronous communication pattern where services send messages to a queue...
MEV Burn: MEV burn is a proposed design that would destroy Maximal Extractable Value instead of paying it to...
Microservices: Microservices is an architecture that splits an app into independent services, each owning one...
Mining Pool: A mining pool is a group of miners who combine hash power, share work, and split block rewards....
Mixture of Experts (MoE): Mixture of Experts is a neural network architecture where only a fraction of the model's parameters...
Model Context Protocol (MCP): Model Context Protocol is an open standard developed by Anthropic that defines how AI models...
Model Distillation: Model distillation is a technique for creating a smaller, faster model that approximates the...
Model Quantization: Model quantization stores weights in lower precision, such as 8-bit integers instead of 32-bit...
Modular Blockchain: A modular blockchain abandons the monolithic architecture where a single network attempts to handle...
Monorepo: A monorepo keeps many projects, packages, or services in one version-control repository instead of...
Multi-Agent Systems: Multi-Agent Systems, MAS, are groups of autonomous software entities that share an environment and...
Multimodal AI: Multimodal AI is a category of models that process multiple data types, text, images, audio, video,...
Named Entity Recognition: Named Entity Recognition finds and classifies people, organizations, locations, dates, money, and...
Neural Network: A neural network is a computational system loosely inspired by biological neurons, consisting of...
Non-Fungible Token (NFT): An NFT is a unique cryptographic token that records ownership of a specific digital or physical...
OAuth: OAuth is an authorization framework that allows third-party applications to access user resources...
Object Detection: Object detection is a computer vision task that both labels objects in an image and locates them...
Optical Character Recognition: OCR converts images of text, scans, signs, and handwriting into machine-readable characters. Older...
Optimistic Rollup: Optimistic rollups treat transactions as valid by default and post state to Layer 1 without an...
Oracle: An oracle is a service that brings off-chain data to smart contracts. Blockchains are closed...
Order Book: An order book is a live list of buy and sell orders for an asset, with size at each price. Bids are...
Overfitting: Overfitting is when a model memorizes the training set instead of learning patterns that hold on...
Parallelized EVM: A parallelized EVM runs Ethereum contract execution across CPU cores instead of one instruction at...
Pattern Recognition: Pattern recognition finds regularities in data, from faces in a crowd to models sorting spam from...
Payment Channel: A payment channel is a Layer 2 method that lets two parties make unlimited instant payments by...
Peer-to-Peer (P2P): Peer-to-Peer (P2P) is a network architecture where each node acts as both client and server,...
Perpetual Futures: Perpetual futures are derivatives that track an asset's price with no expiry. BitMEX popularized...
Perplexity: Perplexity is a metric measuring how well a language model predicts a test dataset, calculated as...
Perplexity Trap: The perplexity trap is the dangerous assumption that lower perplexity on benchmark datasets...
Positional Encoding: Positional encoding is a technique for injecting sequence position information into transformer...
Predictive Analytics: Predictive analytics turns past and present data into forecasts using statistics, data mining, and...
Prefix Tuning: Prefix tuning adapts a frozen language model by learning continuous vectors prepended to the input....
Prompt Engineering: Prompt engineering is the craft of designing inputs to AI systems to produce desired outputs. It...
Prompt Injection: Prompt injection is a security attack where malicious instructions are embedded in content that an...
Prompt Tuning: Prompt tuning learns continuous soft prompt embeddings prepended to inputs while all model weights...
Proof of History (PoH): Proof of History, PoH, is a cryptographic clock that orders events without nodes arguing over...
Proof of Stake: Proof of Stake is consensus where validators propose and attest to blocks based on how much...
Proof of Work: Proof of Work is the original blockchain consensus method, used by Bitcoin. Participants spend...
Protocol Buffer: Protocol Buffers (protobuf) is a language-neutral, platform-neutral data serialization format...
Protocol Revenue: Protocol revenue is the income a DeFi protocol generates from fees and services, representing...
Public-Key Cryptography: Public-key cryptography uses two linked keys: a public key anyone can see and a private key only...
Quantization: Quantization is a model compression technique that reduces the precision of a neural network's...
Quantum Computing: Quantum computing uses qubits and entanglement to run algorithms that do not map cleanly onto...
Question Answering: Question Answering systems answer natural language questions by extracting a span or generating a...
Random Forest: Random Forest is an ensemble that trains many decision trees and combines their predictions. Each...
Rate Limiting: Rate limiting is controlling the number of requests a client can make to a service in a time...
Re-ranking: Re-ranking is a two-stage retrieval pattern: a fast retriever proposes candidates, then a slower,...
Real World Assets (RWA): Real World Assets, RWAs, are tokens that represent off-chain assets such as real estate, treasury...
Reasoning Model: A reasoning model spends time thinking before it answers. It uses an internal chain of thought to...
Recurrent Neural Network: A recurrent neural network processes sequential data by maintaining a hidden state that carries...
Reinforcement Learning: Reinforcement Learning trains an agent through actions and numeric rewards rather than a labeled...
Reinforcement Learning from Human Feedback (RLHF): Reinforcement Learning from Human Feedback is a training technique where humans rank model outputs...
Residual Connection: A residual connection, also called a skip connection, adds a layer's input directly to its output...
REST API: A REST API is an HTTP interface where each resource has a URL and clients act on it with standard...
Restaking: Restaking lets ETH stakers reuse staked ETH as security for protocols beyond Ethereum. Normally,...
Retrieval-Augmented Generation (RAG): Retrieval-Augmented Generation is a technique where an LLM queries external knowledge bases before...
Reward Model: A reward model is a neural network trained to predict human preferences, serving as the learned...
Rollup: A rollup is a Layer 2 that executes transactions off-chain for speed and cost, then posts data or...
Rug Pull: A rug pull is a crypto scam where the team takes the money and leaves holders with a dead token....
SAM: SAM is a Meta AI segmentation model that can mask objects from a point, box, or text prompt without...
Scaffold: A scaffold in AI prompting is a structured template that guides models through complex reasoning by...
Scaling Law: Scaling laws are empirical relationships that predict how neural network performance improves as...
Seed Phrase: A seed phrase, also called a recovery phrase or mnemonic phrase, is a sequence of 12 or 24 randomly...
Self-Attention: Self-attention is an attention mechanism where a sequence attends to itself, allowing each position...
Self-Sovereign Identity (SSI): Self-sovereign identity is a model where you hold your own credentials instead of renting an...
Semantic Similarity: Semantic similarity scores how close two texts are in meaning, not just shared words, for duplicate...
Semi-Supervised Learning: Semi-supervised learning builds models from both labeled and unlabeled data. Fully supervised...
Sentiment Analysis: Sentiment analysis scores the opinion or emotional tone in text, usually as positive, negative, or...
Sequencer: A sequencer orders and batches rollup transactions before posting them to Layer 1. It decides...
Serverless: Serverless is a cloud model where the provider runs the machines. You write functions. They scale...
Serverless Architecture: Serverless architecture means you deploy functions and the cloud runs them on events, billed by...
Sharding: Sharding is a horizontal scaling technique that partitions data across multiple independent...
Slashing: Slashing is an automatic penalty in Proof of Stake. If a validator breaks the protocol, the chain...
Slippage: Slippage is the gap between the price you saw when you started a trade and the price you got. It is...
Smart Contract: A smart contract is self-executing code deployed on a blockchain. It enforces an agreement when the...
Social Recovery: Social recovery restores a wallet through trusted contacts, guardians, instead of a seed phrase....
Soft Fork: A soft fork tightens rules in a way old software still accepts. New blocks remain valid to old...
Softmax: Softmax is a mathematical function that transforms a vector of arbitrary real numbers into a...
Soulbound Token (SBT): Soulbound Tokens (SBTs) are non-transferable, identity-centric digital tokens that uniquely...
Sparse Expert: A sparse expert is one specialist network inside a Mixture of Experts model. Only a few experts run...
Spatial Computing: Spatial computing maps a room and places digital objects in that geometry so software can sit on...
Speculative Decoding: Speculative decoding lets a small draft model propose tokens that a larger target model verifies in...
Speech Recognition: Speech recognition, or ASR, converts spoken language into text for voice interfaces, transcripts,...
SSL/TLS: SSL (Secure Sockets Layer) and TLS (Transport Layer Security) are cryptographic protocols that...
Stablecoin: A stablecoin is a cryptocurrency pegged to a stable asset such as the US dollar. Bitcoin moved from...
State Channel: A state channel lets parties update state off-chain with on-chain security, settling on-chain only...
Supervised Learning: Supervised learning trains a model on labeled examples where every input has a known correct...
Supply Rate: Supply rate is the interest lenders earn for depositing into a DeFi lending protocol. It comes from...
Support Vector Machine (SVM): Support Vector Machine finds the widest-margin hyperplane that separates classes, for...
Sybil Attack: A Sybil attack is when one adversary creates thousands or millions of fake identities to grab power...
Synthetic Data: Synthetic data is artificially generated data created to train, test, or evaluate AI systems, as...
TCP/IP: TCP/IP is the protocol pair that addresses hosts, packs data into packets, routes them, and...
Temperature: Temperature is a parameter that controls how random language model output is by scaling the logits...
Text-to-Speech: Text-to-Speech, TTS, synthesizes spoken audio from written text for voice interfaces, audiobooks,...
Throughput: Throughput measures the rate at which a system processes work over time, typically expressed as...
Time-Weighted Average Price: Time-Weighted Average Price, TWAP, is the average price of an asset over a window, weighted by how...
Token: A token is the smallest unit of text that an LLM processes. It's not a word. A word is often...
Token Burn: Token burning permanently removes tokens by sending them to an address with no private key, a burn...
Tokenization: Tokenization in Web3 means representing ownership of a real-world or digital asset as a token on a...
Tokenizer: A tokenizer converts raw text into a sequence of discrete tokens a language model can process,...
Tokenomics: Tokenomics is the economic design of a cryptocurrency or token system, the rules governing how...
Top-k Sampling: Top-k sampling limits the next token to the k most probable items so rare tokens are less likely to...
Top-p Sampling: Top-p sampling, or nucleus sampling, keeps the smallest token set whose probabilities add up to at...
Total Value Locked: Total Value Locked, TVL, is the dollar value of assets deposited in a DeFi protocol. People use it...
Transfer Learning: Transfer learning is a machine learning approach where knowledge gained from training on one task...
Transformer: The Transformer is a neural network architecture introduced in the 2017 paper "Attention Is All You...
Transformer Architecture: The Transformer, introduced in 2017, processes sequences with attention instead of recurrent...
Utilization Rate: Utilization rate is the share of deposited assets currently borrowed. If $80 million is borrowed...
Validator: A validator is a node that proposes and checks new blocks on a Proof of Stake chain. In Proof of...
Value at Risk: Value at Risk, VaR, is the maximum expected loss on a portfolio over a time horizon at a stated...
Variational Autoencoder: A variational autoencoder is a generative model that learns a probabilistic latent space where...
Vector Database: A vector database is a specialized database optimized for storing and searching high-dimensional...
Vesting Schedule: A vesting schedule releases tokens to recipients gradually over time rather than all at once,...
Vision Transformer: Vision Transformer, ViT, splits an image into patches, treats each patch as a token, and runs a...
Vision Transformers (ViT): Vision Transformers, ViT, adapt the transformer, first built for text, to images. Instead of...
Vocabulary: In a language model, the vocabulary is the full set of tokens the model knows. Each token has an...
Wallet: A wallet is software that stores private keys and lets you use a blockchain. The private key is the...
Web2: Web2 is the read-write web that took over in the early 2000s: sites as platforms and users as...
WebAssembly (Wasm): WebAssembly is a binary instruction format for a stack machine that runs at near-native speed in...
Webhook: A webhook is an HTTP callback: the source POSTs to your URL when an event happens, instead of you...
WebSocket: WebSocket is a communication protocol that provides full-duplex, persistent connections between...
Word Embeddings: Word embeddings are numeric vectors for words that place meaning and relationships in a continuous...
Wrapped Token: A wrapped token is a cryptocurrency that stands in for another asset on a different chain, pegged...
Yield: Yield in DeFi is the return from putting capital into protocols, usually quoted as APY, the...
Yield Farming: Yield farming is deploying crypto across DeFi protocols to stack trading fees, interest, and token...
Zero-Knowledge EVM (zkEVM): A zkEVM is a zero-knowledge rollup that can prove standard Ethereum contract execution. A ZK-Rollup...
Zero-Knowledge Proof: A zero-knowledge proof is a cryptographic method to prove knowledge of information without...
Zero-Knowledge Proofs (ZKPs): Zero-knowledge proofs (ZKPs) are cryptographic protocols that let one party (the prover) convince...
Zero-Shot Learning: Zero-shot learning is when an AI system performs tasks it was never explicitly trained on, using...
ZK Rollup: ZK rollups post cryptographic validity proofs to Layer 1 that the batch is correct, so they skip...

Showing 287 of 287 terms